Full-automatic patrol unmanned aerial vehicle patrol system based on digital twinning
Through digital twin technology and drone organization modules and other means, the problems of insufficient control of drone patrol range and target object occlusion are solved, and more efficient patrol path planning and image analysis accuracy are achieved.
Patent Information
- Application Number
- CN202410962664.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-18
- Publication Date
- 2024-11-08
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing drone patrol technology lacks control over the drone patrol range, resulting in unreasonable patrol planning and affecting efficiency. In addition, during the patrol process, the target object may be blocked, resulting in the inability to make accurate image analysis.
Using a fully automated patrol system based on digital twins, the patrol area is copied through the digital twin module, patrol tasks are generated, image pre-acquisition and coordinate positioning are carried out, and the module is organized and the path is formed using the drone to assign tasks and plan the path, and the task is assigned and planned, and the The occlusion processing module recognizes and adjusts the shooting position to obtain an unobstructed target image.
Effectively control the drone patrol range, ensure reasonable task allocation, reduce patrol time, ensure the accuracy of image analysis, and avoid image analysis errors caused by occlusion.
Smart Images

Figure CN118915781A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of drone technology, and in particular to a fully automated drone inspection system based on digital twins. Background Art
[0002] As the current industry continues to grow, many operating environments require regular inspections to ensure that equipment is operating normally. The coordinated inspection model of drones and manual inspections is not only a trend in technological development, but can also improve inspection quality, efficiency, and benefits.
[0003] The existing drone inspection technology has insufficient control over the drone inspection range, resulting in unreasonable drone inspection planning and affecting the efficiency of drone inspection. At the same time, during the drone inspection process, the target object may be obscured, resulting in the inability to make accurate image analysis. Summary of the invention
[0004] In order to solve the above-mentioned technical problems, a fully automated patrol drone inspection system based on digital twins is provided. This technical solution solves the problem that the existing drone inspection technology proposed in the above-mentioned background technology has insufficient control over the drone inspection range, resulting in unreasonable drone inspection planning and affecting the drone inspection efficiency. At the same time, during the drone inspection process, the target object may be obscured, resulting in the inability to make accurate image analysis.
[0005] In order to achieve the above purpose, the technical solution adopted by the present invention is:
[0006] A fully automated drone inspection system based on digital twins, including:
[0007] A digital twin module, which performs digital twinning on the drone inspection area to form a drone inspection replica area;
[0008] A patrol task generation module, wherein the patrol task generation module generates at least one execution task to be completed when the drone patrols in the drone patrol replication area, wherein the execution task is to acquire an image of a target object;
[0009] An image pre-acquisition module, wherein the image pre-acquisition module pre-acquires a sample image of the target object;
[0010] A task positioning module, wherein the task positioning module performs coordinate positioning on at least one execution task in the drone patrol replication area to obtain at least one target positioning, wherein the target positioning corresponds to the execution task in a one-to-one manner;
[0011] A drone organization module, wherein the drone organization module obtains at least one drone for inspection, organizes and plans the at least one drone, and assigns at least one execution task to the at least one drone;
[0012] A path forming module, wherein the path forming module forms a patrol path for each drone in the drone patrol replication area;
[0013] A patrol module, wherein the patrol module controls the drone to patrol along a patrol path;
[0014] An occlusion processing module, wherein the occlusion recognition module analyzes the actual target image acquired by the drone to determine whether the actual target image is occluded. If not, the actual target image is used as a sample image of the target object. If so, the drone is position-adjusted to obtain at least one target adjustment image, and it is determined whether at least one target adjustment image is occluded. If so, the at least one target adjustment image is synthesized to obtain a target synthesis image, and the target synthesis image is used as the sample image of the target object. If not, the target adjustment image without occlusion is used as the sample image of the target object.
[0015] The abnormality judgment module analyzes the sampled image of the target object, marks the target location corresponding to the target object with abnormality, and subsequently performs maintenance processing at the marked target location.
[0016] Preferably, the digital twinning of the drone inspection area to form a drone inspection replica area includes the following steps:
[0017] Proportional modeling of the drone inspection area is carried out, and a one-to-one replication of the drone inspection area is performed to obtain a digital twin model;
[0018] Reduce the digital twin model by a preset multiple to obtain a replica of the drone inspection area.
[0019] Preferably, generating at least one execution task to be completed during the drone inspection includes the following steps:
[0020] Obtain the task requirements of the drone patrol, and obtain at least one target object that needs to be monitored in the drone patrol replication area;
[0021] An image of the target object is acquired as an execution task to obtain at least one execution task.
[0022] Preferably, the coordinate positioning of at least one execution task to obtain at least one target positioning comprises the following steps:
[0023] Determine the position of the target object corresponding to the execution task in the drone patrol replication area as a feature position;
[0024] The position corresponding to the characteristic position in the drone inspection area is used as the target position;
[0025] The coordinates of the target location are used as the target location.
[0026] Preferably, the organizing and planning at least one drone and assigning at least one execution task to at least one drone comprises the following steps:
[0027] Get the number of at least one drone as the target number;
[0028] Classifying at least one execution task to obtain at least one execution task set, wherein the target positioning intervals of the execution tasks in the execution task set are all less than a preset distance, and the intersections between different execution task sets are empty;
[0029] Obtain the number of at least one execution task set as the feature number;
[0030] When the number of targets is greater than the number of features, the execution task set with the largest number of execution tasks in at least one execution task set is equally divided into two execution task sets, and this step is repeated until the number of targets is equal to the number of features;
[0031] When the number of targets is less than the number of features, merge the execution task set with the least number of execution tasks in at least one execution task set into the execution task set with the smallest distance therefrom, and repeat this step until the number of targets is equal to the number of features, wherein the distance between the two execution task sets refers to the minimum value of the spacing between the target positioning of the execution tasks in the two execution task sets;
[0032] When the number of targets is equal to the number of features, at least one UAV is paired with at least one task set, and the UAV completes the task in the task set.
[0033] Preferably, forming a patrol path for each drone in the drone patrol replication area comprises the following steps:
[0034] Obtain at least one path that traverses all execution tasks in the execution task set as a feature path;
[0035] The shortest characteristic path is obtained as the inspection path of the UAV corresponding to the execution task set.
[0036] Preferably, the determining whether the actual target image is blocked comprises the following steps:
[0037] Acquire a sample image corresponding to the same target object as the actual target image as a feature sample image;
[0038] Performing region division on the actual target image to obtain at least one actual partition;
[0039] Performing region division on the characteristic sample image to obtain at least one first sample partition;
[0040] Calculate the mean pixel value of the pixel points in the actual partition to obtain the actual pixel value mean;
[0041] Calculate the pixel value mean of the pixel points in the first sample partition to obtain the first sample pixel value mean;
[0042] Matching a first sample partition having the smallest difference between the mean value of the first sample pixel values in the at least one first sample partition and the mean value of the actual pixel values in the actual partition to the actual partition;
[0043] Obtain the proportion of the actual partition in the actual target image as the actual proportion;
[0044] Obtaining the proportion of the first sample partition in the feature sample image as the first sample proportion;
[0045] It is determined whether the actual proportion of the actual partition is less than the first sample proportion corresponding to the first sample partition by less than a preset value. If so, there is no occlusion in the actual target image; if not, there is occlusion in the actual target image.
[0046] Preferably, the step of adjusting the position of the drone to obtain at least one target adjustment image comprises the following steps:
[0047] Select at least one shooting point evenly around the target object;
[0048] The drone is controlled to acquire an image of a target object at at least one shooting point to obtain at least one target adjusted image.
[0049] Preferably, the determining whether at least one target adjustment image is blocked comprises the following steps:
[0050] Acquire a sample image of the same target object as the target adjustment image as the target sample image;
[0051] Performing region division on the target adjustment image to obtain at least one target partition;
[0052] Dividing the target sample image into regions to obtain at least one second sample partition;
[0053] Calculate the mean pixel value of the pixels in the target partition to obtain the target pixel value mean;
[0054] Calculate the mean pixel value of the pixel points in the second sample partition to obtain the second sample pixel value mean;
[0055] Matching a second sample partition having the smallest difference between the mean value of the second sample pixel values in at least one second sample partition and the mean value of the target pixel values in the target partition to the target partition;
[0056] Obtain the proportion of the target partition in the target adjusted image as the target proportion;
[0057] Obtaining the proportion of the second sample partition in the target sample image as the second sample proportion;
[0058] It is determined whether the target proportions of the target partitions are all smaller than the second sample proportions corresponding to the second sample partitions than a preset value. If so, there is no occlusion in the target adjusted image; if not, there is occlusion in the target adjusted image.
[0059] Preferably, the step of synthesizing at least one target adjustment image to obtain a target synthesized image comprises the following steps:
[0060] Acquire at least one target partition corresponding to the same second sample partition as a calibration target partition;
[0061] Select a calibration target partition whose target proportion and the second sample proportion of the second sample partition have a difference smaller than a preset value as the final target partition;
[0062] The final target partitions are merged according to the arrangement positions of the second sample partitions in the target sample image to obtain a target comprehensive image.
[0063] Compared with the prior art, the present invention has the following beneficial effects:
[0064] By setting up the drone organization module, path formation module and occlusion processing module, the drone patrol range can be allocated to ensure that the distances between the multiple execution tasks assigned to each drone are within a reasonable range, thereby limiting the patrol time of each drone and avoiding some drones from patrolling too long paths. At the same time, by changing the shooting position and image processing methods, images of unobstructed targets can be obtained, and more accurate image analysis can be made based on this. BRIEF DESCRIPTION OF THE DRAWINGS
[0065] Figure 1 It is a structural schematic diagram of a fully automated inspection drone inspection system based on digital twins of the present invention;
[0066] Figure 2 A schematic diagram of the process of performing digital twinning of the drone inspection area to form a drone inspection replica area according to the present invention;
[0067] Figure 3 A schematic diagram of a process for generating at least one execution task to be completed during a drone inspection according to the present invention;
[0068] Figure 4 A schematic diagram of a process of performing coordinate positioning on at least one execution task and obtaining at least one target positioning according to the present invention;
[0069] Figure 5 A schematic diagram of a process of organizing and planning at least one unmanned aerial vehicle and assigning at least one execution task to at least one unmanned aerial vehicle according to the present invention;
[0070] Figure 6 A schematic diagram of a process of forming a patrol path for each drone in a drone patrol replication area according to the present invention;
[0071] Figure 7 It is a schematic diagram of a flow chart of judging whether an actual target image is blocked according to the present invention;
[0072] Figure 8 A schematic diagram of a process of adjusting the position of a drone to obtain at least one target adjustment image according to the present invention;
[0073] Fig. 9 A schematic diagram of a flow chart of determining whether at least one target adjustment image is blocked according to the present invention;
[0074] Fig.10 The present invention is a flow chart of synthesizing at least one target adjustment image to obtain a target synthesized image. DETAILED DESCRIPTION
[0075] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are only examples, and those skilled in the art may think of other obvious variations.
[0076] Reference Figure 1 As shown in the figure, a fully automated drone inspection system based on digital twins includes:
[0077] A digital twin module, which performs digital twinning on the drone inspection area to form a drone inspection replica area;
[0078] A patrol task generation module, wherein the patrol task generation module generates at least one execution task to be completed when the drone patrols in the drone patrol replication area, wherein the execution task is to acquire an image of a target object;
[0079] An image pre-acquisition module, wherein the image pre-acquisition module pre-acquires a sample image of the target object;
[0080] A task positioning module, wherein the task positioning module performs coordinate positioning on at least one execution task in the drone patrol replication area to obtain at least one target positioning, wherein the target positioning corresponds to the execution task in a one-to-one manner;
[0081] A drone organization module, wherein the drone organization module obtains at least one drone for inspection, organizes and plans the at least one drone, and assigns at least one execution task to the at least one drone;
[0082] A path forming module, wherein the path forming module forms a patrol path for each drone in the drone patrol replication area;
[0083] A patrol module, wherein the patrol module controls the drone to patrol along a patrol path;
[0084] An occlusion processing module, wherein the occlusion recognition module analyzes the actual target image acquired by the drone to determine whether the actual target image is occluded. If not, the actual target image is used as a sample image of the target object. If so, the drone is position-adjusted to obtain at least one target adjustment image, and it is determined whether at least one target adjustment image is occluded. If so, the at least one target adjustment image is synthesized to obtain a target synthesis image, and the target synthesis image is used as the sample image of the target object. If not, the target adjustment image without occlusion is used as the sample image of the target object.
[0085] The abnormality judgment module analyzes the sampled image of the target object, marks the target location corresponding to the target object with abnormality, and subsequently performs maintenance processing at the marked target location.
[0086] Reference Figure 2 As shown in the figure, the digital twin of the drone inspection area and the formation of the drone inspection replica area include the following steps:
[0087] Proportional modeling of the drone inspection area is carried out, and a one-to-one replication of the drone inspection area is performed to obtain a digital twin model;
[0088] Reduce the digital twin model by a preset multiple to obtain a replica of the drone inspection area.
[0089] Organizational planning is carried out in the drone inspection replication area. Since the drone inspection replication area is a reduced one-to-one model, it is relatively convenient to organize and plan the drones, and the analysis data can be intuitively displayed in the drone inspection replication area.
[0090] Reference Figure 3 As shown, generating at least one execution task to be completed when a drone inspection is performed includes the following steps:
[0091] Obtain the task requirements of the drone patrol, and obtain at least one target object that needs to be monitored in the drone patrol replication area;
[0092] An image of the target object is acquired as an execution task to obtain at least one execution task.
[0093] When drones patrol, they need to complete given patrol tasks. The rationality of task allocation directly affects the efficiency of drone patrol. When the allocation is unreasonable, some drone patrol paths are too short and some are too long, which results in the overall patrol time being prolonged. Therefore, it is necessary to design a task allocation model for task allocation.
[0094] Reference Figure 4 As shown, performing coordinate positioning on at least one execution task to obtain at least one target positioning includes the following steps:
[0095] Determine the position of the target object corresponding to the execution task in the drone patrol replication area as a feature position;
[0096] The position corresponding to the characteristic position in the drone inspection area is used as the target position;
[0097] The coordinates of the target location are used as the target location.
[0098] The allocation of execution tasks mainly depends on coordinate positioning. The target positioning distance that the UAV needs to complete the execution task is the minimum path distance when the UAV patrols. When the distance between the two target positionings is too large, the flight time of the UAV will increase. Therefore, when planning the execution of the task, it is necessary to obtain the target positioning of the execution task, and then plan the path based on this.
[0099] Reference Figure 5 As shown, organizing and planning at least one drone and assigning at least one execution task to at least one drone includes the following steps:
[0100] Get the number of at least one drone as the target number;
[0101] Classifying at least one execution task to obtain at least one execution task set, wherein the target positioning intervals of the execution tasks in the execution task set are all less than a preset distance, and the intersections between different execution task sets are empty;
[0102] Obtain the number of at least one execution task set as the feature number;
[0103] When the number of targets is greater than the number of features, the execution task set with the largest number of execution tasks in at least one execution task set is equally divided into two execution task sets, and this step is repeated until the number of targets is equal to the number of features;
[0104] When the number of targets is less than the number of features, merge the execution task set with the least number of execution tasks in at least one execution task set into the execution task set with the smallest distance therefrom, and repeat this step until the number of targets is equal to the number of features, wherein the distance between the two execution task sets refers to the minimum value of the spacing between the target positioning of the execution tasks in the two execution task sets;
[0105] When the number of targets is equal to the number of features, at least one UAV is paired with at least one task set, and the UAV completes the task in the task set.
[0106] There is a correlation between planning and the number of drones. When allocating, each drone is assigned tasks evenly as much as possible. However, due to the positional relationship of the tasks, forced even distribution may cause some drones to obtain tasks that are far apart. In this solution, when allocating, tasks with similar distances are divided into task sets. As a result, the number of task sets obtained may deviate from the number of drones. Therefore, it is impossible to directly correspond the task sets to drones one-to-one. Therefore, it is necessary to process the task sets so that the number of processed task sets corresponds to the drones one-to-one. When processing, it is necessary to ensure that each drone obtains as few tasks as possible, thereby limiting its patrol time.
[0107] Reference Figure 6 As shown, forming a patrol path for each drone in the drone patrol replication area includes the following steps:
[0108] Obtain at least one path that traverses all execution tasks in the execution task set as a feature path;
[0109] The shortest characteristic path is obtained as the inspection path of the UAV corresponding to the execution task set.
[0110] Reference Figure 7 As shown, judging whether the actual target image is blocked includes the following steps:
[0111] Acquire a sample image corresponding to the same target object as the actual target image as a feature sample image;
[0112] Performing region division on the actual target image to obtain at least one actual partition;
[0113] Performing region division on the characteristic sample image to obtain at least one first sample partition;
[0114] Calculate the mean pixel value of the pixel points in the actual partition to obtain the actual pixel value mean;
[0115] Calculate the pixel value mean of the pixel points in the first sample partition to obtain the first sample pixel value mean;
[0116] Matching a first sample partition having the smallest difference between the mean value of the first sample pixel values in the at least one first sample partition and the mean value of the actual pixel values in the actual partition to the actual partition;
[0117] Obtain the proportion of the actual partition in the actual target image as the actual proportion;
[0118] Obtaining the proportion of the first sample partition in the feature sample image as the first sample proportion;
[0119] It is determined whether the actual proportion of the actual partition is less than the first sample proportion corresponding to the first sample partition by less than a preset value. If so, there is no occlusion in the actual target image; if not, there is occlusion in the actual target image.
[0120] Since shooting may occur at different angles, the imaging ratios will be different, and it is impossible to directly compare images to determine whether there is occlusion. In this solution, the shooting angles are different, and the imaging changes of each part of the target are consistent. Therefore, the proportion of each part remains unchanged. When there is no occlusion, the proportion of each part is consistent with the proportion of each part in the feature sample image. Therefore, it can be used to determine whether there is occlusion.
[0121] Reference Figure 8 As shown, adjusting the position of the drone to obtain at least one target adjustment image includes the following steps:
[0122] Select at least one shooting point evenly around the target object;
[0123] The drone is controlled to acquire an image of a target object at at least one shooting point to obtain at least one target adjusted image.
[0124] The occlusion situation will change when the drone takes pictures at different angles. There may be an angle where the target object is not occluded, and this image is used for analysis. However, occlusion may still occur at each angle. Therefore, at least one target adjustment image is integrated to obtain a target integrated image without occlusion, and the target integrated image is used for analysis.
[0125] Reference Fig. 9 As shown, judging whether at least one target adjustment image is blocked includes the following steps:
[0126] Acquire a sample image of the same target object as the target adjustment image as the target sample image;
[0127] Performing region division on the target adjustment image to obtain at least one target partition;
[0128] Dividing the target sample image into regions to obtain at least one second sample partition;
[0129] Calculate the mean pixel value of the pixels in the target partition to obtain the target pixel value mean;
[0130] Calculate the mean pixel value of the pixel points in the second sample partition to obtain the second sample pixel value mean;
[0131] Matching a second sample partition having the smallest difference between the mean value of the second sample pixel values in at least one second sample partition and the mean value of the target pixel values in the target partition to the target partition;
[0132] Obtain the proportion of the target partition in the target adjusted image as the target proportion;
[0133] Obtaining the proportion of the second sample partition in the target sample image as the second sample proportion;
[0134] It is determined whether the target proportions of the target partitions are all smaller than the second sample proportions corresponding to the second sample partitions than a preset value. If so, there is no occlusion in the target adjusted image; if not, there is occlusion in the target adjusted image.
[0135] Reference Fig.10 As shown, synthesizing at least one target adjustment image to obtain a target synthesized image includes the following steps:
[0136] Acquire at least one target partition corresponding to the same second sample partition as a calibration target partition;
[0137] Select a calibration target partition whose target proportion and the second sample proportion of the second sample partition have a difference smaller than a preset value as the final target partition;
[0138] The final target partitions are merged according to the arrangement positions of the second sample partitions in the target sample image to obtain a target comprehensive image.
[0139] The occlusion conditions are different at different shooting angles. Therefore, the unoccluded parts of the images at multiple angles can be integrated to obtain the target integrated image.
[0140] Furthermore, the present solution also proposes a storage medium on which a computer-readable program is stored. When the computer-readable program is called, the above-mentioned fully automated patrol drone inspection system based on digital twins is executed.
[0141] It is understandable that the storage medium may be a magnetic medium, such as a floppy disk, a hard disk, or a magnetic tape; an optical medium, such as a DVD; or a semiconductor medium, such as a solid state drive (SSD).
[0142] To sum up, the advantages of the present invention are: by setting up a drone organization module, a path formation module and an occlusion processing module, the patrol range of the drone can be allocated to ensure that the distances between the multiple execution tasks assigned to each drone are within a reasonable range, thereby limiting the patrol time of each drone and avoiding that the patrol paths of some drones are too long. At the same time, by changing the shooting position and the image processing method, the image of the target object that is not obstructed can be obtained, and based on this, a more accurate image analysis can be made.
[0143] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions only describe the principles of the present invention. The present invention may be subject to various changes and improvements without departing from the spirit and scope of the present invention. These changes and improvements fall within the scope of the present invention. The scope of protection claimed by the present invention is defined by the attached claims and their equivalents.
Claims
1. A fully automated drone inspection system based on digital twins, characterized in that: include: A digital twin module, which performs digital twinning on the drone inspection area to form a drone inspection replica area; A patrol task generation module, wherein the patrol task generation module generates at least one execution task to be completed when the drone patrols in the drone patrol replication area, wherein the execution task is to acquire an image of a target object; An image pre-acquisition module, wherein the image pre-acquisition module pre-acquires a sample image of the target object; A task positioning module, wherein the task positioning module performs coordinate positioning on at least one execution task in the drone patrol replication area to obtain at least one target positioning, wherein the target positioning corresponds to the execution task in a one-to-one manner; A drone organization module, wherein the drone organization module obtains at least one drone for inspection, organizes and plans the at least one drone, and assigns at least one execution task to the at least one drone; A path forming module, wherein the path forming module forms a patrol path for each drone in the drone patrol replication area; A patrol module, wherein the patrol module controls the drone to patrol along a patrol path; An occlusion processing module, wherein the occlusion recognition module analyzes the actual target image acquired by the drone to determine whether the actual target image is occluded. If not, the actual target image is used as a sample image of the target object. If so, the drone is position-adjusted to obtain at least one target adjustment image, and it is determined whether at least one target adjustment image is occluded. If so, the at least one target adjustment image is synthesized to obtain a target synthesis image, and the target synthesis image is used as the sample image of the target object. If not, the target adjustment image without occlusion is used as the sample image of the target object. The abnormality judgment module analyzes the sampled image of the target object, marks the target location corresponding to the target object with abnormality, and subsequently performs maintenance processing at the marked target location.
2. According to the digital twin-based fully automated inspection drone inspection system of claim 1, it is characterized in that: The digital twinning of the drone inspection area to form a drone inspection replica area includes the following steps: Proportional modeling of the drone inspection area is carried out, and a one-to-one replication of the drone inspection area is performed to obtain a digital twin model; Reduce the digital twin model by a preset multiple to obtain a replica of the drone inspection area.
3. According to the digital twin-based fully automated inspection drone inspection system of claim 2, it is characterized in that: The generating of at least one execution task to be completed during the drone inspection comprises the following steps: Obtain the task requirements of the drone patrol, and obtain at least one target object that needs to be monitored in the drone patrol replication area; An image of the target object is acquired as an execution task to obtain at least one execution task.
4. According to the digital twin-based fully automated inspection drone inspection system of claim 3, it is characterized in that: The coordinate positioning of at least one execution task to obtain at least one target positioning comprises the following steps: Determine the position of the target object corresponding to the execution task in the drone patrol replication area as a feature position; The position corresponding to the characteristic position in the drone inspection area is used as the target position; The coordinates of the target location are used as the target location.
5. According to the digital twin-based fully automated inspection drone inspection system of claim 4, it is characterized in that: The organizing and planning of at least one UAV and assigning at least one execution task to at least one UAV comprises the following steps: Get the number of at least one drone as the target number; Classifying at least one execution task to obtain at least one execution task set, wherein the target positioning intervals of the execution tasks in the execution task set are all less than a preset distance, and the intersections between different execution task sets are empty; Obtain the number of at least one execution task set as the feature number; When the number of targets is greater than the number of features, the execution task set with the largest number of execution tasks in at least one execution task set is equally divided into two execution task sets, and this step is repeated until the number of targets is equal to the number of features; When the number of targets is less than the number of features, merge the execution task set with the least number of execution tasks in at least one execution task set into the execution task set with the smallest distance therefrom, and repeat this step until the number of targets is equal to the number of features, wherein the distance between the two execution task sets refers to the minimum value of the spacing between the target positioning of the execution tasks in the two execution task sets; When the number of targets is equal to the number of features, at least one UAV is paired with at least one task set, and the UAV completes the task in the task set.
6. The fully automated inspection drone inspection system based on digital twin according to claim 5 is characterized in that: The forming of a patrol path for each drone in the drone patrol replication area comprises the following steps: Obtain at least one path that traverses all execution tasks in the execution task set as a feature path; The shortest characteristic path is obtained as the inspection path of the UAV corresponding to the execution task set.
7. The fully automated inspection drone inspection system based on digital twin according to claim 6 is characterized in that: The method of determining whether the actual target image is blocked comprises the following steps: Acquire a sample image corresponding to the same target object as the actual target image as a feature sample image; Performing region division on the actual target image to obtain at least one actual partition; Performing region division on the characteristic sample image to obtain at least one first sample partition; Calculate the mean pixel value of the pixel points in the actual partition to obtain the actual pixel value mean; Calculate the pixel value mean of the pixel points in the first sample partition to obtain the first sample pixel value mean; Matching a first sample partition having the smallest difference between the mean value of the first sample pixel values in the at least one first sample partition and the mean value of the actual pixel values in the actual partition to the actual partition; Obtain the proportion of the actual partition in the actual target image as the actual proportion; Obtaining the proportion of the first sample partition in the feature sample image as the first sample proportion; It is determined whether the actual proportion of the actual partition is less than the first sample proportion corresponding to the first sample partition by less than a preset value. If so, there is no occlusion in the actual target image; if not, there is occlusion in the actual target image.
8. The fully automated inspection drone inspection system based on digital twin according to claim 7 is characterized in that: The step of adjusting the position of the drone to obtain at least one target adjustment image comprises the following steps: Select at least one shooting point evenly around the target object; The drone is controlled to acquire an image of a target object at at least one shooting point to obtain at least one target adjusted image.
9. The fully automated inspection drone inspection system based on digital twin according to claim 8 is characterized in that: The step of determining whether at least one target adjustment image is blocked comprises the following steps: Acquire a sample image of the same target object as the target adjustment image as the target sample image; Performing region division on the target adjustment image to obtain at least one target partition; Dividing the target sample image into regions to obtain at least one second sample partition; Calculate the mean pixel value of the pixels in the target partition to obtain the target pixel value mean; Calculate the mean pixel value of the pixel points in the second sample partition to obtain the second sample pixel value mean; Matching a second sample partition having the smallest difference between the mean value of the second sample pixel values in at least one second sample partition and the mean value of the target pixel values in the target partition to the target partition; Obtain the proportion of the target partition in the target adjusted image as the target proportion; Obtaining the proportion of the second sample partition in the target sample image as the second sample proportion; It is determined whether the target proportions of the target partitions are all smaller than the second sample proportions corresponding to the second sample partitions than a preset value. If so, there is no occlusion in the target adjusted image; if not, there is occlusion in the target adjusted image.
10. The fully automated inspection drone inspection system based on digital twin according to claim 9 is characterized in that: The step of synthesizing at least one target adjustment image to obtain a target synthesized image comprises the following steps: Acquire at least one target partition corresponding to the same second sample partition as a calibration target partition; Select a calibration target partition whose target proportion and the second sample proportion of the second sample partition have a difference smaller than a preset value as the final target partition; The final target partitions are merged according to the arrangement positions of the second sample partitions in the target sample image to obtain a target comprehensive image.
Citation Information
Cited By
Communication line inspection path planning method based on unmanned aerial vehicle
CN119937600A
